Cardiac Arrhythmia Modeling With Animal ECG Data Sets

ISEF Category: Animal Sciences

Subcategory: Physiology  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

A tiny shift in heart-cell timing can turn a steady beat into a wild rhythm change. The FitzHugh-Nagumo equations let you model that timing with a small set of numbers instead of a full biology lab. Public ECG datasets from animals let you test whether your model actually matches real signals, not just pretty plots.

What Is It?

The FitzHugh-Nagumo equations are a simple math model for electrical activity in a heart cell. Think of them like a push and a reset. One variable rises fast when the cell fires, and the other brings it back down, which helps you mimic spikes and recovery without modeling every ion channel.

An ECG, or electrocardiogram, is the trace that shows how the heart's electrical signal moves through time. Animal ECG records on PhysioNet give you real waveforms to compare against your simulation. You can ask whether your model captures beat timing, spike shape, and rhythm changes that look like arrhythmia.

Why This Is a Good Topic

This is a strong science fair topic because you can change model parameters, measure the effect, and compare the output against public data. It connects to heart rhythm research, veterinary physiology, and signal analysis. You can do the work with free data and a laptop, and you will learn how to handle differential equations, data cleaning, and model scoring.

Research Questions

• How does changing the FitzHugh-Nagumo recovery parameter alter simulated beat timing? ?
• What is the effect of adding noise on the model's match to animal ECG shape? ?
• Does one parameter set fit one species better than another in public PhysioNet records? ?
• To what extent can the model reproduce RR interval variability seen in arrhythmic records? ?
• Which ECG features, such as spike width, peak timing, or baseline shift, are best matched by the model? ?
• How does a simple pacing or perturbation term change the model's ability to create irregular beats? ?

Basic Materials

• Laptop or desktop computer.
• Python 3 with NumPy, SciPy, pandas, and matplotlib.
• Jupyter Notebook or Google Colab.
• Free access to PhysioNet ECG datasets.
• Spreadsheet software for quick checks.

Advanced Materials

• High-performance workstation for repeated simulations.
• Python with NumPy, SciPy, pandas, matplotlib, and scikit-learn.
• MATLAB or Julia for alternate model fitting.
• WFDB Python Package or WFDB software for reading PhysioNet records.
• Git for version control and experiment tracking.
• Access to a university server or cluster for batch runs and sensitivity tests.

Software & Tools

• Python: Solves the differential equations and analyzes ECG features.
• Jupyter Notebook: Keeps code, plots, and notes in one place.
• WFDB Python Package: Reads PhysioNet waveform records and annotations.
• SciPy: Fits parameters and solves the model numerically.
• matplotlib: Plots model output and real ECG traces on the same scale.

Experiment Steps

1. Define the ECG feature or arrhythmia pattern you want the model to match.
2. Choose one animal dataset from PhysioNet and decide how you will clean and split the records.
3. Set up the FitzHugh-Nagumo equations and decide which parameters you will fit, hold fixed, or sweep.
4. Build a comparison metric so you can score how close each simulation is to the real ECG features.
5. Plan controls that test whether improvements come from the model or from overfitting the dataset.
6. Decide how you will report uncertainty, sensitivity, and limits of the model.

Common Pitfalls

• Fitting the model to a single record and treating the result as general, which hides how species and leads differ.
• Comparing simulated output to raw ECG data before resampling, which makes timing error look smaller or larger than it really is.
• Changing several parameters at once, which makes it impossible to tell which setting caused the rhythm shift.
• Using only visual overlap between traces, which misses beat-to-beat timing errors and outlier spikes.
• Mixing training and test records from the same animal, which inflates the apparent accuracy of your model.

What Makes This Competitive

A stronger version does more than fit one animal ECG trace. It tests whether one parameter set can predict different records, species, or arrhythmia patterns, then backs that up with clear error metrics and sensitivity checks. You can raise the bar by comparing several optimization methods, separating training and test records, and showing where the model breaks. That kind of analysis shows real understanding of both the physiology and the math.

Project Variations

• Compare mouse, rat, and rabbit ECG records to see whether one parameter set transfers across species.
• Add a pacing or noise term and test whether irregular beats match the data better.
• Compare FitzHugh-Nagumo fits with a simpler threshold model to see which one tracks timing and spike shape more closely.

Learn More

• PhysioNet: Search the PhysioNet site for open ECG databases, waveform tools, and annotation notes.
• NIH PubMed: Search review articles on cardiac electrophysiology, arrhythmia modeling, and ECG analysis.
• MIT OpenCourseWare: Look for differential equations and computational biology lecture notes for model setup.
• WFDB Software Package: Find documentation for reading PhysioNet records and annotations on the PhysioNet software pages.
• PLOS Computational Biology: Search open articles on heart rhythm models and signal analysis.